as suggested by the openldap administrators guide, the preferred method to kill slapd is with an INT signal. Another method could lead to database corruption/data loss. http://www.openldap.org/doc/admin22/runningslapd.html#Stopping%20slapd The init script that comes with net-nds/openldap-2.1.30-r2, (its is possibly the same with other versions) Kills the daemon with a TERM signal. I myself had database corruption a couple of times allready, I cannot blame this, because my box was under strange conditions every time it happend. But i cannot rule it out either. Reproducible: Couldn't Reproduce Steps to Reproduce: 1. Reboot (last time i had quite a few runaway processes [configure crashed], couldnt kill them, ps/top just hang while they tried to list the processes) Actual Results: The database was corrupted. i had to remove it and install from a backup ldif. Expected Results: Not cause database corruption I dont know if this was the cause, and maybe there is a good reason for the TERM signal i'am not aware of. This is my first bug(suggestion), so please bare with me. I tried to be as clear as possible, but its probably some ranting you have to struggle through..
forgot the solution.. change the init script stop function to this: start-stop-daemon -s 2 --stop --quiet --pidfile /var/run/openldap/slapd.pid
in cvs now.